Lightnews — Scholar-powered news

LightNews

Mike Frank

@mcxfrank.bsky.social

9.5K followers 930 following 340 posts

Cognitive scientist at Stanford. Open science advocate. Symbolic Systems Program director. Bluegrass picker, slow runner, dad. http://langcog.stanford.edu

langcog.stanford.edu

Posts Media Videos Starter Packs

Reposted by Mike Frank

Dr Mircea Zloteanu ☀️ 🌊🌴 @mzloteanu.bsky.social · 9h

#statstab #434 Exposing omitted moderators: Explaining why effect sizes differ in the social sciences

Thoughts: Maybe our models are too simple to makdle the generalisable claims we want.

#bias #moderator #causalinference #heterogeneity #effect

doi.org/10.1073/pnas...

Exposing omitted moderators: Explaining why effect sizes differ in the social sciences | PNAS

Policymakers increasingly rely on behavioral science in response to global challenges, such as climate change or global health crises. But applicat...

Reposted by Mike Frank

Cameron Ellis @camerontellis.bsky.social · 8h

Astronaut meme: "Wait, it's all perception?" "Always has been"

Reposted by Mike Frank

Dawei Bai @daweibai.bsky.social · 12h

Happy to share that our BBS target article has been accepted: “Core Perception”: Re-imagining Precocious Reasoning as Sophisticated Perceiving
With Alon Hafri, @veroniqueizard.bsky.social, @chazfirestone.bsky.social & Brent Strickland
Read it here: doi.org/10.1017/S014...
A short thread [1/5]👇

Mike Frank @mcxfrank.bsky.social · 8h

Science fiction science, by the ever-thoughtful Iyad Rahwan: www.nature.com/articles/s41...

The science fiction science method - Nature

The ‘science fiction science’ method simulates future technologies and collects quantitative data on the attitudes and behaviours of participants in various future scenarios, with the aim of predicting impacts of future technologies before they arrive.

Mike Frank @mcxfrank.bsky.social · 13h

How do you measure early language development in languages where you don't have word lists?

New paper: Measuring children’s early vocabulary in low-resource languages using a Swadesh-style word list - led by @kachergis.bsky.social and @alvinwmtan.bsky.social

osf.io/njm7d_v1

Title and abstract of the paper

Reposted by Mike Frank

Nora Newcombe @noranewcombe.bsky.social · 2d

Thrilled to see this paper out! It's the culmination of a project begun in the depths of the pandemic with Sabrina Karjack and @zoengo.bsky.social . We continue our exploration of how children generalize when their episodic memory is not yet mature.
www.nature.com/articles/s41...

The dependence of children’s generalization on episodic memory varies with age and level of abstraction - Nature Communications

Children’s ability to generalize from episodic memories varies by both age and the level of abstraction. Here, the authors show that lower level generalization increasingly depends on episodic memory with age, whereas higher level generalization shows no such relationship.

Reposted by Mike Frank

Sasha Boguraev @ COLM @sashaboguraev.bsky.social · May 27

A key hypothesis in the history of linguistics is that different constructions share underlying structure. We take advantage of recent advances in mechanistic interpretability to test this hypothesis in Language Models.

New work with @kmahowald.bsky.social and @cgpotts.bsky.social!

🧵👇!

Mike Frank @mcxfrank.bsky.social · 8d

One big question is, start with songs or start with rudiments/scales/exercises! Holistic top down or bottom up foundations.

In college teaching, we have to figure out whether we want to start with the history/foundations/principles compared with compelling examples (“songs”).

Mike Frank @mcxfrank.bsky.social · 8d

In contrast, the worst have learning goals like “expose to X” or “show Y”… they’re a bit like pedagogical concerts but not like real teaching.

Mike Frank @mcxfrank.bsky.social · 8d

For example, I loved a class on playing clawhammer guitar. It was an hour long and the teacher made you do the basic clawhammer motion on the guitar literally for the entire hour to solidify a complicated motor routine. All the rest was gravy because you left having put in that practice…

Mike Frank @mcxfrank.bsky.social · 8d

The best have really clear and immediate learning goals. I liked your AC/DC example because some of the best classes I’ve taken define a learning goal that matters and take immediate steps to get you moving on it in class time.

Mike Frank @mcxfrank.bsky.social · 8d

At various times I’ve studied mandolin, guitar, bass, cello, and voice. Right now mostly mando and voice. Bluegrass and folk music has a very robust group teaching tradition (eg in camps and festivals). Often these have very uneven pedagogy.

Mike Frank @mcxfrank.bsky.social · 8d

I play music a lot and study with many music teachers. :)

Mike Frank @mcxfrank.bsky.social · 8d

Great analogy! I think a lot about transfer between music teaching and university teaching (and vice versa).

Debbie Gale Mitchell @heydebigale.bsky.social · 8d

Something about learning the drums as a middle-aged lady:

It’s boosting my empathy for my students in my chemistry classes. 🥁🧪

I already play the piano, but even still— learning a new instrument requires so much practice and getting comfortable with not being good at something.

Picture of a white woman with green hair wearing white headphones holding drum sticks in front of a drum set.

Mike Frank @mcxfrank.bsky.social · 9d

oh great - thank you!!

Mike Frank @mcxfrank.bsky.social · 9d

Bonus: full text of the paper "The Yerkes-Dodson Law Repealed" pasted below:

There is a flaw
In the evidence for the Yerkes-Dodson Law.
To call it “ubiquitous”
Is pretty iniquitous.

journals.sagepub.com/doi/abs/10.2...

Mike Frank @mcxfrank.bsky.social · 9d

Anyone know what the best recent evidence on the Yerkes-Dodson law is? Thanks in advance!

Mike Frank @mcxfrank.bsky.social · 10d

Thank you!!

Reposted by Mike Frank

Rebecca Saxe @rebeccasaxe.bsky.social · 10d

Totally agree with Mikes description of this project as a wild journey, utterly joyous true collaboration, and satisfying first step for quantitative predictive rational model of habituation.

Not the first time I’ve suggested a “first step” in research that required a whole PhD to complete. 😉

Mike Frank @mcxfrank.bsky.social · 10d

Ever wonder how habituation works? Here's our attempt to understand:

A stimulus-computable rational model of visual habituation in infants and adults doi.org/10.7554/eLif...

This is the thesis of two wonderful students: @anjiecao.bsky.social @galraz.bsky.social, w/ @rebeccasaxe.bsky.social

infant data from experiment 1

conceptual schema for different habituation models

title page

results from experiment 2 with adults

Mike Frank @mcxfrank.bsky.social · 10d

This project used literally every trick I know about measurement, experimental design, and modeling (and I learned many more along the way by following Anjie Gal and Rebecca). I hope it looks clear and obvious from the paper but it was a wild voyage of discovery!

Mike Frank @mcxfrank.bsky.social · 10d

Featuring:
- our best attempt at a within-subjects habituation design in young infants!
- a Bayesian model fit to neural net representations
- a shocking amount of GPU compute to do approximate inference in what should be a simple multi-level model

Mike Frank @mcxfrank.bsky.social · 10d

Ever wonder how habituation works? Here's our attempt to understand:

A stimulus-computable rational model of visual habituation in infants and adults doi.org/10.7554/eLif...

This is the thesis of two wonderful students: @anjiecao.bsky.social @galraz.bsky.social, w/ @rebeccasaxe.bsky.social

infant data from experiment 1

conceptual schema for different habituation models

title page

results from experiment 2 with adults

Mike Frank @mcxfrank.bsky.social · 13d

GPT or BERT, why not both? arxiv.org/pdf/2410.24159

Winner of second BabyLM competition uses a clever and hilariously simple hack - align BERT and GPT so you can use both. Seems to be very efficient for learning from child-scale data.

Reposted by Mike Frank

Monica Ellwood-Lowe @mellwoodlowe.bsky.social · 24d

I’m hiring!! 🎉 Looking for a full-time Lab Manager to help launch the Minds, Experiences, and Language Lab at Stanford. We’ll use all-day language recording, eye tracking, & neuroimaging to study how kids & families navigate unequal structural constraints. Please share:
phxc1b.rfer.us/STANFORDWcqUYo

Research Coordinator, Minds, Experiences, and Language Lab in Graduate School of Education, Stanford, California, United States

The Stanford Graduate School of Education (GSE) seeks a full-time Research Coordinator (acting lab manager) to help launch and coordinate the Minds,.....

Mike Frank @mcxfrank.bsky.social · 24d

A first blogpost from the LEVANTE team, introducing our global project perspective.

A good intro if you're interested in learning more about cross-cultural developmental data collection using LEVANTE.

levante-network.org/global-colla...

Global Collaboration to Capture Variability in Child Development - Levante

Ensuring that children around the world can flourish is one of the most important goals of our time. Developmental science has made major strides in understanding how children grow, learn, and thrive ...

levante-network.org